fix(i18n): compile fuzzy translations into the backend .mo files (#41648)

Co-authored-by: Amin Ghadersohi <amin.ghadersohi@gmail.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
Evan Rusackas
2026-07-02 10:43:13 -07:00
committed by GitHub
parent 6c2c814b5c
commit aa40934e7f
2 changed files with 16 additions and 3 deletions

View File

@@ -332,15 +332,28 @@ cd superset-frontend
npm run build-translation
# Backend
pybabel compile -d superset/translations
pybabel compile --use-fuzzy -d superset/translations
```
`--use-fuzzy` includes `#, fuzzy` entries in the compiled `.mo` files. Superset
serves fuzzy translations on purpose: the frontend build (`po2json --fuzzy`)
already includes them, `flask fab babel-compile` (used by the release images)
compiles with `-f`, and the production `Dockerfile` compiles with `--use-fuzzy`
as well. This keeps machine-generated (and other draft) translations visible in
the UI rather than falling back to English while they await review.
### Backfilling missing translations with AI
For languages with many untranslated strings, the repo includes a script that
uses Claude AI to generate draft translations for any missing entries. All
AI-generated strings are marked `#, fuzzy` and tagged with an attribution
comment so that human reviewers know they need to be checked before merging.
comment so that human reviewers know they need to be checked.
Note that `#, fuzzy` marks a translation as *needing review*, not as *withheld*:
both the frontend and backend builds serve fuzzy entries (see [Applying
translations](#applying-translations) above), so an AI-generated string is shown
in the UI as soon as it is built and deployed. Reviewers should verify each
entry and remove the `#, fuzzy` flag to promote it to a confirmed translation.
#### Prerequisites